This analysis was conducted within a cross-sectional population-based biomarker survey of non-pregnant, non-lactating women 15–40 years of age, in Andhra Pradesh, India (NCT04048330). The study design and findings have been previously described (48, 53, 54). Briefly, trained nurse enumerators conducted a census among all households (rural: 3,124; urban: 3,428) within the catchment area of the research site, Arogyavaram Medical Centre (AMC). All rural households (n=3,124) and a simple random sample of urban households (n=1,000) were revisited for screening and recruitment. One woman (15–40 y, not pregnant, not lactating) per household was selected using an a priori algorithm developed by a biostatistician at the U.S. Centers for Disease Control and Prevention (CDC; CER). Study staff referred women who were currently pregnant or had severe anemia (Hb<8.0 g/dL) to a local clinic for standard of care.

The study protocol (NCT04048330) was approved by the Institutional Review Board at Cornell University and the Institutional Ethics Committees at St. John’s Research Institute and AMC; and received clearance from the Indian Council of Medical Research Health Ministry’s Screening Committee. The protocol was reviewed in accordance with the U.S. Centers for Disease Control and Prevention (CDC) human research protection procedures and was determined to be a non-research, routine surveillance activity. A non-disclosure agreement for personally identifiable information and a data sharing agreement for de-identified data were established. All participants provided written informed consent or assent prior to data collection.

Data were collected at a single study visit at AMC by trained nurse enumerators on electronic tablets (55), and included sociodemographic, clinical, anthropometric, and dietary data. Anthropometric measurements were collected in triplicate and included weight using a calibrated digital scale to the nearest 0.01 kg (Seca 874DR, Hamburg, Germany), height using a stadiometer to the nearest 0.1 cm (Weigh and Measure LLC, Olney, MD); waist and hip circumferences using a retractable, non-stretch measuring tape to the nearest 0.1 cm; mid-upper arm circumference using a 65 cm measuring tape to the nearest 0.1 cm (Weigh and Measure LLC, Olney, MD); and triceps skinfold using a caliper to the nearest 0.5 mm (Lange Skinfold Caliper; Beta Technology, Santa Cruz, CA). Bioelectrical impedance analysis (BIA) was used to evaluate whole body (WF%) and trunk (TF%) percent fat mass in women ≥18 y (BC-418 MA; Tanita Corporation, Tokyo, Japan; 8 electrode). Diastolic blood pressure (DBP) and systolic blood pressure (SBP) were measured in triplicate using a manual sphygmomanometer.

Venous blood was collected, processed, and stored using standardized protocols (53). Hemoglobin (Hb) and glycated hemoglobin (HbA1c) were evaluated in real-time via Coulter Counter (Coulter HMX) and nephelometry (Agappe Diagnostics, Kerala, India), respectively. Vitamin B₁₂ biomarkers were analyzed in archived samples (≤−80°C) in batch after the completion of data collection, including serum total vitamin B₁₂ via chemiluminescence (E411Roche Diagnostics, Mannheim, Germany), plasma methylmalonic acid (MMA) and homocysteine (Hcy) via gas chromatography-mass spectrometry (GCMS-SQ, 5975, Agilent Technologies, CA USA), and plasma holotranscobalamin (holoTC) via an enzyme-immunoassay (ELISA, Abbexa). Red blood cell (RBC) folate and serum folate concentrations were measured via the World Health Organization (WHO)-recommended microbiologic assay (56). Ranges and mean coefficients of variation for quality controls for each assay were: vitamin B₁₂ (pmol/L; low: 218–328, 7.5%, medium: 238–439, 6.1%, high: 448–615, 4.5%), MMA (μmol/L; pooled: 0.41–0.52, 3.7%), Hcy (μmol/L; pooled: 6.82–8.10, 2.8%), RBC folate (nmol/L; low: 127–300, 8.8%; high: 322–595, 7.5%), and serum folate (nmol/L; low: 5.6–11.3, 7.4%; high: 20.6–41.9, 7.1%).

Vitamin B₁₂ was categorized as <148 and <221 pmol/L (57, 58). Elevated MMA, a functional biomarker of vitamin B₁₂ status, was defined as >0.26 μmol/L (3, 58). Both circulating and functional biomarkers are recommended for assessing vitamin B₁₂ status (3, 57, 58); impaired vitamin B₁₂ status was defined as vitamin B₁₂ <148 pmol/L and MMA >0.26 μmol/L. Other cutoffs for lower vitamin B₁₂ status were also evaluated (MMA >0.26 and >0.37 μmol/L (3, 58); holoTC <35 and <40 pmol/L (3, 58); combined vitamin B₁₂ indicator, cB₁₂ ≤ −1.5 (59); vitamin B₁₂ ≥148-<221 pmol/L); results were not significant and were not reported. Vitamin B₁₂ status as a three-level variable (<148, ≥148-<221, and ≥221 pmol/L) and quintiles were also considered; results were similar to primary analyses and were not reported. Elevated Hcy was defined as >10.0 and >15.0 μmol/L (60). RBC folate was categorized as <305 and <748 nmol/L, respectively, and serum folate was categorized as <7 nmol/L (60, 61).

Continuous measures of body mass index (BMI; weight [kg] / height [m]²), waist circumference (WC), and waist-hip ratio (WHR) were examined among all women (≥15 y). Categorical measures of BMI, WC, and WHR, which were established for adults, were examined among adult women (≥18 y). Among adult women (≥18 y), BMI was categorized using WHO criteria (i.e., BMI <18.5, 18.5 to <25.0, 25.0 to <30.0, and ≥30.0 kg/m²) (62) for comparability across studies; analyses using BMI cut-offs for Asian populations (63) had similar findings (data not shown). Elevated waist circumference (WC) was defined as >88.9 cm (converted from >35 inches) (64) and elevated WHR was defined as ≥0.85 (65), based on NIH and WHO criteria.

Continuous and categorical measures of WF% and TF%, which were assessed using a BIA model calibrated for ages 18–84 y, were examined among adult women (≥18 y). Elevated WF% and TF% were defined among adult women (≥18 y) as >35%, as suggested by the WHO and American Association of Clinical Endocrinology (66–69).

Among all women (≥15 y), blood pressure was categorized as normal (SBP <120 and DBP <80 mmHg), elevated blood pressure (SBP 120–129 and DBP <80 mmHg), stage 1 hypertension (SBP 130–139 or DBP 80–89 mmHg), and stage 2 hypertension (SBP ≥140 or DBP ≥90 mmHg) (70). Hypertension defined as SBP ≥130 or DBP ≥80 mm Hg (i.e., stage 1 or 2 hypertension) was also considered in analyses. Among all women, HbA1c was categorized as <5.7%, ≥5.7% to <6.5%, and ≥6.5%, using WHO and American Diabetes Association criteria corresponding to cutoffs for normoglycemia, prediabetes, and diabetes, respectively (71, 72); HbA1c ≥5.7% was also considered in analyses.

Continuous variables were categorized using conventional cutoffs, where available. Biomarker values that were below or above the limits of detection (LOD) for assays were set to half the LOD or 2-times the LOD, respectively, and continuous biomarker values were natural logarithmically transformed prior to analyses. Other variables were assessed for skewness and determined to be normally distributed and were not transformed. Continuous variables were presented as geometric means (GM) and 95% confidence intervals (95% CI). Chi-square statistics and one-way ANOVA were used to evaluate differences in categorical and continuous metabolic characteristics, respectively, by rural/urban residence and vitamin B₁₂ status.

Linear and binomial regression models were used to examine the associations of vitamin B₁₂ status (continuous and categorical) with continuous and categorical metabolic characteristics (adiposity, central adiposity, blood pressure, glycated hemoglobin), respectively. Results from models are presented as β estimates and standard error (unit change in continuous metabolic characteristics) and prevalence ratios and 95% confidence intervals (categorical metabolic characteristics). Variables were evaluated as potential effect measure modifiers (e.g., rural/urban residence, parity, BMI ≥25.0 kg/m² categories, folate status); effect modification was not detected (p-value for interaction term <0.10). Potential confounders were evaluated using the approach recommended by Rothman and Greenland (73), in which risk factors for an outcome (i.e., factors associated with an outcome in prior literature) that led to a >10% change in estimate were included in models. The following variables were considered for inclusion in multivariable models: age, education, currently married, gravidity, employment status. Results were similar when considering rural/urban residence and dietary variables (vegetarian dietary preference and frequency of animal-source food consumption) for inclusion in multivariable models; therefore, parsimonious models without these variables are presented. Potential nonlinear associations of vitamin B₁₂ status with metabolic characteristics were investigated using restricted cubic splines and assessed using the likelihood ratio test, comparing models with only the linear term to models with both the linear and cubic spline terms (74). Observations with missing data for covariates (<3% per variable, i.e., education, n=5; marital status, n=5; gravidity, n=5; employment status, n=28) were retained in analyses using the missing indicator method (i.e., in which a dummy variable to indicate missing and a variable with missing values set to the reference value were included in the model) (75). The findings for the complete case analysis were similar to those from the analysis using the missing indicator method; the results based on the missing indicator method (i.e., retaining missing data in the analysis) are reported. Sensitivity analyses were conducted among adults (≥18 y), and among women with BMI ≥18.5 kg/m².

Statistical analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA). Analyses were reproduced by the Cornell Center for Social Sciences (CCSS) Data Archive and Replication Team.

Characteristics of women in this study are provided in Table 1 and Supplemental Table 1, and a flow chart of study participants is presented in Supplemental Figure 1. Most women had some formal schooling (83.8%), were currently married (79.4%), and had 4 to 5 family members in their household. A total of 93.1% of women reported following a non-vegetarian diet (i.e., consumption of meat, poultry, or fish), although consumption of animal-source foods was low (<25% of women consumed eggs, poultry, meat, or fish ≥1 time per week). A total of 48.3% of women had vitamin B₁₂ concentrations <148 pmol/L, 74.3% had vitamin B₁₂ concentrations <221 pmol/L; and 37.8% of women had impaired vitamin B₁₂ status (i.e., vitamin B₁₂ <148 pmol/L and MMA >0.26 μmol/L).

Metabolic characteristics of the study population are presented in Table 2. A total of 23.4% of adult women (≥18 y) had overweight (BMI ≥25.0 to <30.0 kg/m²) and 9.6% had obesity (BMI ≥30.0 kg/m²). A total of 13.4% of adult women had elevated WC (>88.9 cm), 20.8% had elevated WHR (≥0.85), 39.3% had elevated WF% (>35%), and 36.2% had elevated TF% (>35%). In the overall population, one-fourth of women had HbA1c ≥5.7% (HbA1c ≥5.7 to <6.5%: 20.0%; HbA1c ≥6.5%: 5.0%); and 18.6% had hypertension (Stage 1: SBP 130–139 or DBP 80–89 mmHg: 16.4%; Stage 2 (SBP ≥140 or DBP ≥90 mmHg: 2.2%). Women living in urban settings had higher BMI and WC compared to women in rural areas (p<0.05), and adult women living in urban settings had higher WF% compared to adult women in rural areas (p<0.05).

Metabolic characteristics by vitamin B₁₂ strata (<148 pmol/L vs. ≥148 pmol/L) are presented in Figures 1 and 2. Women with vitamin B₁₂ concentrations <148 pmol/L had higher BMI and WC compared to women with vitamin B₁₂ ≥148 pmol/L (p<0.05). Adult women with vitamin B₁₂ concentrations <148 pmol/L had higher WF% and TF%, and higher prevalence of overweight or obesity, and elevated WC, WHR, WF%, and TF%, compared to adult women with vitamin B₁₂ ≥148 pmol/L (p<0.05) (Supplemental Table 2). Other metabolic characteristics, including HbA1c, blood pressure, and hypertension, did not differ by vitamin B₁₂ <148 pmol/L. Results were similar when examining metabolic characteristics across three levels of vitamin B₁₂ status (<148, ≥148-<221, and ≥221 pmol/L; Supplemental Table 3).

The associations between continuous total vitamin B₁₂ concentrations and metabolic characteristics are presented in Table 3. Nonlinear associations were not detected; continuous vitamin B₁₂ concentrations were associated with metabolic characteristics consistently across strata of vitamin B₁₂ status (Supplemental Figure 2). Higher continuous vitamin B₁₂ concentrations were associated with lower indicators of central adiposity and overweight including WC (β: −1.64; standard error [SE]: 0.61; p=0.007), in multivariable models adjusting for age, education, currently married, gravidity, and employment, and BMI (β [SE]: −0.92 [0.27]; p=0.001), in multivariable models adjusting for age, currently married, gravidity, and employment; and lower WF% (β [SE]: −1.34 [0.49]; p=0.007) and TF% (β [SE]: −1.51 [0.61]; p=0.013), in multivariable models among adults adjusting for currently married, gravidity, and employment. Among adult women, higher continuous vitamin B₁₂ concentrations were associated with lower prevalence of overweight (prevalence ratio [PR]: 0.82; 95% confidence interval [95% CI]: 0.71, 0.96; p=0.013), and elevated WC (PR: 0.64 [95% CI: 0.49, 0.85]; p=0.002), WHR (PR: 0.81 [95% CI: 0.66, 1.00]; p=0.049), and WF% (PR: 0.86 [95% CI: 0.76, 0.98]; p=0.028) in multivariable models. Continuous total vitamin B₁₂ concentrations were not associated with HbA1c, elevated HbA1c, systolic or diastolic blood pressure, or hypertension.

The associations between vitamin B₁₂ <148 pmol/L and metabolic characteristics are presented in Table 4. Women with vitamin B₁₂ concentrations <148 pmol/L had 1.29 kg/m² higher BMI (β [SE]: 1.29 [0.32]; p<0.0001) in multivariable models adjusting for age, currently married, and employment, and higher WC (β [SE]: 2.65 [0.72]; p<0.001) in multivariable models adjusting for age, currently married, gravidity, and employment. Adults with vitamin B₁₂ concentrations <148 pmol/L had higher WF% (β [SE]: 1.72 [0.60]; p=0.004) and TF% (β [SE]: 1.99 [0.73]; p=0.007) compared to adult women with vitamin B₁₂ ≥148 pmol/L. Similarly, in adult women, vitamin B₁₂ concentrations <148 pmol/L was associated with increased prevalence of overweight (PR: 1.31 [95% CI: 1.09, 1.58]; p=0.004), and elevated WC (PR: 1.85 [95% CI: 1.32, 2.60]; p<0.001), WHR (PR: 1.38 [95% CI: 1.07, 1.78], p=0.013), WF% (PR: 1.29 [95% CI: 1.10, 1.51]; p=0.001), and TF% (PR: 1.25 [95% CI: 1.06, 1.49]; p=0.010), compared to vitamin B₁₂ ≥148 pmol/L in multivariable models. Vitamin B₁₂ <148 pmol/L was not significantly associated with other metabolic characteristics, including HbA1c, blood pressure, and hypertension.

The associations between impaired vitamin B₁₂ status (vitamin B₁₂ <148 pmol/L and MMA >0.26 μmol/L) and metabolic characteristics are presented in Table 5. Impaired vitamin B₁₂ status was associated with higher BMI (β [SE]: 0.73 [0.34]; p=0.031) and higher WF% (≥18 y; β [SE]: 1.28 [0.61]; p=0.036) and TF% (≥18 y; β [SE]: 1.51 [0.76]; p=0.046) in multivariable models. Adult women with impaired vitamin B₁₂ status had higher prevalence of overweight (PR: 1.22 [95% CI: 1.02, 1.46]; p=0.032), elevated WC (PR: 1.48 [95% CI: 1.07, 2.06]; p=0.020), and elevated TF% (PR: 1.20 [95% CI: 1.02, 1.42]; p=0.033) in multivariable models. Impaired vitamin B₁₂ status was not significantly associated with other metabolic characteristics, including HbA1c, blood pressure, or hypertension.

The associations between vitamin B₁₂ <221 pmol/L and metabolic characteristics are presented in Supplemental Table 4. Women with vitamin B₁₂ <221 pmol/L had higher BMI (β [SE]: 1.04 [0.37]; p=0.006) and WC (β [SE]: 2.09 [0.83]; p=0.012). Adult women with vitamin B₁₂ <221 pmol/L had higher WF% (β [SE]: 1.59 [0.67]; p=0.018), and TF% (β [SE]: 1.88 [0.83]; p=0.024), and higher prevalence of elevated WC (PR: 1.54 [95% CI: 1.00, 2.36]; p=0.037), compared to adult women with vitamin B₁₂ ≥221 pmol/L in multivariable models. Lower vitamin B₁₂ status defined using other biomarkers or cutoffs (MMA >0.37 μmol/L; holoTC <35 and <40 pmol/L; vitamin B₁₂ ≥148-<221 pmol/L; combined vitamin B₁₂ indicator, cB₁₂ ≤ −1.5) were not associated with indicators of adiposity, HbA1c, or blood pressure (data not shown). Vitamin B₁₂ status as a three-level variable (<148, ≥148-<221, and ≥221 pmol/L) and quintiles were similar to primary analyses (data not shown).

Findings from all sensitivity analyses (among women ≥18 y, Supplemental Table 5; and among women with BMI ≥18.5 kg/m², data not shown) were similar to the full population.

The prevalence of overweight and obesity (78, 81), diabetes (76, 77, 82), and hypertension (83, 84) have increased globally. In this population, the prevalence of overweight in adult women (≥18 y; BMI≥25 kg/m²: 33.0%) was similar to state-level estimates in women 15–49 y in India (NFHS-5; Andhra Pradesh: 36.3%) (85, 86) and in other population-based studies in Southern India (≥18 y: 28.4–42.2%) (87, 88). In contrast, the prevalence of central adiposity among adult women in the current study (WC>88.9 cm: 13.4%; WHR≥0.85: 20.8%) was lower than state-level estimates (WC>88.0 cm: 28.4%; WHR≥0.85: 48.8%) (85, 86); findings from other population-based studies have varied (87–92). Similar to the current study, state-level estimates for overweight and elevated waist circumference were higher for women living in urban settings compared to rural settings (86). The prevalence of elevated HbA1c corresponding to cutoffs for prediabetes (≥5.7%: 25.0%) and diabetes (≥6.5%: 4.9%) screening in women ≥15 y in this study was lower than other population-based studies in women in West Bengal (≥18 y; ≥5.7%: 34.8%; ≥6.5%: 9.7%) (93) and Mysore (30–59 y; ≥5.7%: 57.9%) (91), India. The use of different indicators for diabetes screening in other studies in Southern India (85, 86, 88, 89, 94–96), including fasting blood glucose (≥126 mg/dL; 1.1–12.4%), which measures different aspects of glucose metabolism, constrains the comparability of findings (71).

In the current study, lower vitamin B₁₂ status was associated with increased central adiposity (elevated WC, WHR, TF%), overweight, and elevated WF%. To date, observational studies on vitamin B₁₂ and adiposity have largely focused on BMI, and findings have been heterogeneous. Lower total vitamin B₁₂ status (continuous or <150 pmol/L) was associated with higher BMI in in adult women with a high prevalence of vitamin B₁₂ deficiency (<148 or <150 pmol/L; 20–50%) in India and Turkey (97, 98). In clinical (23, 26, 99, 100) and population-based studies (38, 39) in adult men and women in India, findings have been heterogeneous: vitamin B₁₂ <140 pmol/L was associated with elevated WC but not overweight (n=1,500) (38), and vitamin B₁₂ concentrations were not associated with overweight, or elevated WC or WHR (n=1,374) (39).

In population-based studies in adults in settings with lower prevalence of vitamin B₁₂ deficiency (Vietnam, U.S., Canada, Denmark), findings have been heterogeneous: lower vitamin B₁₂ status was associated with higher BMI, WC, WF% (37), and odds of obesity (41, 101) in some studies, but not others (40, 43). In analyses of U.S. NHANES data, lower vitamin B₁₂ quartiles were associated with higher odds of obesity in men and women (2011–2014) (41), while overweight was not associated with odds of vitamin B₁₂ <221 pmol/L in premenopausal women (BMI ≥18 kg/m²; 1988–1994) (43). Population characteristics (e.g., vitamin B₁₂ status, sex, genetics (37, 102, 103)) may contribute to inconsistent findings. The prevalence of severe obesity is rising in the U.S. (104), and the prevalence of obesity is expected to double from 2020 to 2035 in low- and middle-income countries (78); additional population-based data across regions is important as global risk factors and burden change.

In the current study, vitamin B₁₂ status was not associated with HbA1c. Studies to date on vitamin B₁₂ status and HbA1c have primarily been conducted in clinical and obstetric populations in other settings (Peru, Poland, Israel, Saudi Arabia) with heterogenous results: lower vitamin B₁₂ status was associated with higher HbA1c in some studies (22, 105, 106) but not others (107, 108). In studies in adults in India, lower vitamin B₁₂ status was associated with higher fasting glucose (n=75) (26) and insulin resistance (n=300) (23) in clinical populations; higher insulin resistance and risk of type 2 diabetes in postpartum women (n=795) (98); and higher odds of type 2 diabetes in a nested case-control study (n=1,500) (38). Lower vitamin B₁₂ status was not associated with fasting glucose (109) or insulin resistance in women in other settings (n=122–346) (97, 110). The lack of association in the current study and the heterogenous results in the literature could be due to the use of different biomarkers (e.g., HbA1c vs. fasting glucose) and populations at different risk for developing diabetes (e.g., women of reproductive age vs. patients with metabolic syndrome or postpartum women). For example, HbA1c may be affected by factors outside of glucose metabolism, such as iron deficiency (111) which is prevalent in this population (>40%) (54). Approximately 50% of diabetes cases are undiagnosed globally, with the highest burden in India and other low- and middle-income countries (77); identifying risk factors could help inform screening strategies.

In the current study, vitamin B₁₂ status was not associated with blood pressure or hypertension. There is limited data on vitamin B₁₂ and hypertension, and findings are heterogeneous. In India, vitamin B₁₂ concentrations were lower in adults with hypertension compared to normotensive individuals (n=130) (99), but vitamin B₁₂ concentrations were not associated with blood pressure (100) or hypertension (23) in other clinical populations (n=300–321). Lower vitamin B₁₂ status has been associated with other cardiovascular outcomes, including higher triglycerides (23, 26, 100, 109, 112), which were not evaluated in this study.

Women with vitamin B₁₂ concentrations <221 pmol/L had increased adiposity, compared to ≥221 pmol/L, while results did not differ between women with vitamin B₁₂ concentrations ≥148-<221 pmol/L compared to women with concentrations ≥221 pmol/L. Analyses were constrained by the substantial proportion of women with vitamin B₁₂ concentrations <148 pmol/L (48.3%) compared to ≥148-<221 pmol/L (26.1%). In the U.S., the prevalence of vitamin B₁₂ ≥148-<221 pmol/L is more common than vitamin B₁₂ <148 pmol/L among non-pregnant, non-lactating women (11% vs. 2%) (1, 113, 114), as well as in non-pregnant women in other countries, including Canada (20–79 y; 20% vs. 6%) (101), the UK (19–39 y; 44% 150–258 pmol/L vs. 4% <150 pmol/L) (115), Ecuador (12–49 y; 20% vs. 4%) (116), Belize (15–49 y; 33% vs. 17%) (117), Guatemala (15–49 y; 30 % vs. 20%) (118), Columbia (18–49 y; 18.9% vs. 18.5%) (119), Malawi (15–49 y; 30% vs. 12%) (120), and Ghana (15–49 y; 12% vs. 7%) (121). Although results are similar to analyses conducted in the U.S. (41), additional data is needed to inform the role of vitamin B₁₂ in metabolic health in the context of concentrations ≥148-<221 pmol/L, given its global prevalence.

Both circulating and functional biomarkers are recommended for assessing vitamin B₁₂ status (3, 57, 58). Findings for impaired vitamin B₁₂ status (vitamin B₁₂<148 pmol/L and MMA>0.26 μmol) were similar to vitamin B₁₂ <148 pmol/L, and demonstrate a robust association between vitamin B₁₂ <148 pmol/L and adiposity in this population. Vitamin B₁₂ metabolism is closely linked to other nutrients, such as folate (60). Lower serum or plasma folate and higher RBC folate concentrations have been reported in individuals with overweight and obesity (122–127). In the current analysis, findings did not vary by RBC or serum folate status, consistent with other studies (98, 128) and with data suggesting that underlying conditions such as poor renal function and vitamin B₁₂ malabsorption may explain adverse health outcomes related to the combination of high folate and low vitamin B₁₂ status (129, 130).

This population-based study is among the largest to date to examine associations of vitamin B₁₂ with metabolic characteristics in women. This analysis included circulating and functional biomarkers of vitamin B₁₂ status, and comprehensive assessment of body composition—in a population with a high prevalence of vitamin B₁₂ concentrations <148 pmol/L and adverse metabolic characteristics. This study has several limitations. The cross-sectional design does not inform temporal or causal relationships or changes over time. The limited range of concentrations within the lower quality control is a limitation of the total vitamin B₁₂ assay. Non-fasted archived blood samples constrained the analysis of other indicators of glycemic control. Although several potential confounders were evaluated, findings may be influenced by unmeasured or residual confounders. The response rate in the parent study (54%) is comparable to other population-based biomarker surveys (131); although participants were similar to non-participants (48), unmeasured factors may have influenced participation.